Patients and Methods
Details of the design of the multi-element study of bone health and body composition have been published.⁴ From a cohort of children and adolescents who were diagnosed with ALL more than a decade earlier at McMaster Children’s Hospital (MCH) a study sample was constructed (Fig 1). Treatment had been administered on protocols of the Dana Farber Cancer Institute Childhood ALL Consortium⁷ of which MCH had been a member since 1985.
The current report focuses on bone geometry, density and architecture as assessed by pQCT performed with a Stratec XCT 2000 instrument (Stratec Medezintechnik, Pforzheim, Germany) with the Stratec Analysis Software v6.0 (Orthometrix Inc, White Plains, NY, USA). Measurements were made by a single experienced operator at the 4 % sites of the length of the radius and tibia in the non-dominant limbs as well as at the 20% site of the radius and the 38% and 66% sites of the tibia, all with 0.2 mm voxel size and 2.2 mm slice thickness. The 4% (metaphyseal) sites represent trabecular bone and the proximal (diaphyseal) sites cortical bone.⁵ In order to emphasize metrics of high clinical importance and relevance to bone biology and biomechanics, including bone strength, as well as to allow comparisons with published reference values, analyses focused on 4 metrics at the metaphyseal sites – Total bone mineral content (BMC), Total vBMD, Trabecular BMC and Trabecular vBMD – and 6 metrics at the diaphyseal sites – Total BMC, Total vBMD, Cortical BMC, Cortical vBMD, Cortical Thickness and Polar Strength-Strain Index (SSI); a total of 10 selected from the many metrics provided by the Stratec software.
In addition to the standard volumetric density and geometry endpoints derived from the Stratec analysis software, measures of apparent trabecular architecture captured at the 4% metaphyseal sites were derived from the pQCT images. Specifically, the 4% metaphyseal images were post-processed with custom software developed by one of the authors.⁸ This builds on the work of CLG and colleagues, beginning 25 years ago.9-12 As reported,8 the software uses edge detection and thresholding steps to segment the trabecular bone and to quantify its structure. Fig 2 illustrates some of the image processing steps. The connectivity of the trabecular network is based on strut analysis in which the network is considered to consist of one-dimensional struts (Fig 2A).9 A junction between three or more struts is defined as a node. A strut which is connected at one end but is free at the other is called a free end. Struts representing trabeculae running perpendicular to the image appear as points and are deemed to be isolated points. The connectivity of the bone architecture can also be assessed by quantifying the appearance of holes in the network due to loss of trabecular elements (Fig 2B). As more elements are lost, the average hole size reflects the increased mean distance between trabecular struts.11 Overall, from the 78 metrics available from post-processing, 7 were selected with respect to apparent trabecular bone architecture– Trabecular Thickness (mm), Trabecular Number (1/mm), Average Hole Size (mm2), Maximum Hole Size (mm2), Number of Free Ends per Unit Area (Free End Density – mm-2), Number of Isolated Points per Unit Area (Isolated Points Density – mm-2) and Connectivity Index (Number of Nodes per Unit Connected Network). A bone with high connectivity, and so strength (resistance to bending, torsion and compression), has a large number of nodes with few free ends and isolated points as well as small marrow pores (hole sizes). The custom program allows greater sensitivity to detect changes associated with perturbed bone metabolism that we have described in children with ALL.13
Of note, pQCT images can be acquired in a short period of time (10 minutes) with low radiation exposure (approximately 5µSv),14 equivalent to about 2 days of natural radiation. Comparisons were made with published data on healthy subjects data from relevant age groups using Stratec XCT 2000 and 3000 instruments.5,15-23 Comparative values for the 4% radius site are available from 4 publications,15, 18, 19, 22 and there are 4 reports on the 4% tibia site.16,19,21,23 Fewer were found for the diaphyseal sites – 20% radius,19 38% tibia,17,20,21 and 66% tibia.16,23Information from these sources were provided in Supplementary Tables 1 and 2. There is no “gold standard” set of reference values for pQCT in children and adolescents.24 However, reference values for comparison at the 4% radius22 and the 4% and 38% tibia25 are available from a sample of more than 220 healthy Caucasian subjects, allowing the calculation of Z scores at these sites. The normative values for those aged 20-40 years are taken as equivalent to those in healthy 19 year old males and females.26 These data were obtained using Stratec equipment and software very similar to those used in the current study.
In light of the differences in body composition in this cohort according to sex,27 the pQCT metrics were also examined separately in males and females. Data from pQCT are presented as means and standard deviations or as medians and ranges when appropriate. Subgroup analyses were undertaken on the 14 subjects who had received a bisphosphonate during therapy to correct osteoporosis, defined as an LS BMD Z score < -2.0.28,29 Parametric data were compared with 2 tailed t tests and non-parametric data by chi-squared tests. Correlations were measured with Pearson’s correlation coefficient for parametric data and with Spearman’s rho for non-parametric data. Comparisons of groups who did and did not receive a bisphosphonate were performed by Fisher’s exact test. All analyses were performed with SPSS, version 24.
This study was approved by the Hamilton Integrated Research Ethics Board (project 10-508-5) which represents Hamilton Health Sciences, McMaster University and St. Joseph’s Healthcare Hamilton. Informed consent was obtained from each participant.